Skip to content

Share the gallery's look fixes and answer marks with published patterns - #54

Merged
aharlap merged 9 commits into
mainfrom
feat/shared-look
Oct 23, 2026
Merged

aharlap merged 9 commits into
mainfrom
feat/shared-look

Conversation

@aharlap

@aharlap aharlap commented Oct 23, 2026 •

Copy link
Copy Markdown
Contributor

Brings the review gallery's shared fixes and owner decisions (2026-10-22) into the published patterns, before gallery blocks are ported.

  • Headings inside a pattern use the body face. On praxity.io, h2 and h3 titles in six patterns picked up the site's display face.
  • lib/base.css gains the gallery's fixes: select arrow off the border, a 3 px focus ring, a neutral dashed mark for a correct answer the learner didn't choose, synthesised italics, no hover on locked choices, and no rule directly under the scene header.
  • One set of answer marks: Correct / Incorrect / Correct answer (French: Correct / Incorrect / Bonne réponse). This replaces "Not quite", "Missed" and highlight's own words.
  • The scene header holds one title. dont-know's scoring line moves under it.
  • AGENTS.md "Look" states these rules.

Not in this PR: the conversation header's role line (needs an owner call), a round focus ring on bare radios (none exist in published patterns), and shared mark helpers in lib/ (next slice).

Checks: npm test (411), typecheck, types:check, budget, demo; the touched browser specs in Chromium (527 passed). Full three-engine run in CI.

Review: 0/2/6 0/1/0; fixes 2. Opus at d6bd3de, Sol at b70ed28; fixes in 736d0a7, 91cd2f5, b70ed28 and a188a0d (highlight keeps its five old string keys as deprecated aliases for Studio).

🤖 Generated with Claude Code

aharlap and others added 9 commits October 22, 2026 21:24
Hosts often set a display face on h1 to h3 directly (praxity.io uses
Bricolage), which beats the font .lp sets for inheritance. Six live
patterns showed their scene titles in the site's display face. base.css
now sets --lp-font-heading on every heading, scene title and run-in
inside .lp.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The owner closed these in the gallery's shell.css; published patterns
now get them from base.css:

- The select arrow is drawn 14px inside the border in every engine.
  Forced colours keep the native arrow.
- data-lp-mark="missed" draws a neutral dashed row for a correct answer
  the learner didn't choose.
- Keyboard focus is a 3px ring in --lp-focus, with Highlight in forced
  colours, including bare checkboxes and radios outside choice rows.
  Pattern rings move to 3px to match.
- em is synthesised italic when the host sets font-synthesis: none.
- Locked choice rows show no hover or pointer.
- A section right under the scene header draws no second rule.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The scene header holds one title and no subtitle line (owner,
2026-10-22). The scoring rule moves from lp-scene-sub to the first
line under the header.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Owner decision, 2026-10-22. A chosen correct answer reads "Correct", a
chosen wrong one "Incorrect", and a correct answer the learner didn't
choose "Correct answer" (French "Correct", "Incorrect", "Bonne
reponse"). The third mark is neutral and dashed, so it doesn't read as
the learner's pick.

- dont-know and test-out: "Not quite" / "Pas tout a fait" becomes
  "Incorrect", and the unchosen correct row is marked missed.
- highlight: "Key idea", "Answers the question", "Missed", "Not a key
  idea" and the default evidence line give way to the same three
  words, under the same keys. An author's note still follows
  "Incorrect"; a missed target is grey rather than orange.

Explanations stay in the feedback text, not the mark.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Headings in the body face, em synthesis, one title in the scene header,
the three answer marks, the 3px focus ring and the drawn select arrow.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Review round 1. The locked-row hover rule also matched a chosen row
without a mark, such as "I don't know" after Check, and turned its
accent border grey. It now skips checked rows. An invalid select's
2px border reset its padding and ran the label under the drawn arrow;
it now keeps 39px at the end. A forced-colours check covers the dashed
missed row.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Screen readers ran the mark word into the note ("Incorrect You feel
overwhelmed"). A visually hidden full stop after the word adds the
pause.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Praxity Studio still reads correctKey, correctEvidence, wrongKey,
wrongEvidence and missed. Removing them would make its answer list
throw and its verdicts read "undefined" once it bumps its pin. The
keys stay in both languages with the new words ("Correct",
"Incorrect", "Correct answer" / "Bonne reponse"), marked deprecated in
favour of markCorrect, markWrong and markAnswer.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@aharlap
aharlap marked this pull request as ready for review October 23, 2026 03:22
@aharlap
aharlap merged commit 494305e into main Oct 23, 2026
2 checks passed
@aharlap
aharlap deleted the feat/shared-look branch October 23, 2026 03:22
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant